Foundation Models and Deep Learning for Intelligent SAR Image Understanding and Analysis

Dear Colleagues, 

Synthetic aperture radar (SAR) enables reliable Earth observation regardless of illumination conditions, cloud cover, or adverse weather, making it a critical data source for applications such as military reconnaissance, maritime surveillance, disaster assessment, and urban management. In recent years, the volume of SAR data has increased rapidly, driven by missions such as Sentinel-1 and Gaofen-3, as well as emerging and planned satellite constellations. High-resolution, multi-polarization, and multi-mode SAR imagery is now routinely acquired over large geographic areas, substantially expanding both the scale and complexity of interpretation tasks, ranging from target-level detection and recognition to scene-level segmentation and change detection. To address these growing demands, SAR interpretation methods have evolved rapidly over the past decade, shifting from hand-crafted feature engineering toward data-driven deep learning. More recently, foundation models have begun to push SAR interpretation beyond conventional perception toward generation and reasoning. Self-supervised pre-training enables transferable representations to be learned from large volumes of unlabeled SAR data, while generative models support SAR image synthesis, restoration, and enhancement. Meanwhile, vision–language models, together with the agentic systems built upon them, introduce language-based interaction, reasoning capabilities, and autonomous workflows into SAR interpretation. Against this backdrop, how to design, adapt, and rigorously evaluate foundation models in accordance with the distinctive imaging physics and signal characteristics of SAR has emerged as a central research question in remote sensing.

This Special Issue will bring together recent advances in foundation models and deep learning for intelligent SAR image understanding and analysis. Particular emphasis is placed on methods that address the unique characteristics of SAR data while improving the generalization, scalability, and interpretability of intelligent SAR processing. We welcome contributions spanning foundation model pre-training and adaptation, self-supervised and multimodal learning, generative modeling, vision–language modeling, and agent-based SAR interpretation, as well as their applications to target recognition, semantic understanding, change analysis, and large-scale or time-series Earth observation. Studies on SAR-specific datasets, benchmarks, evaluation protocols, and efficient model deployment are also encouraged. By advancing methodologies for the processing, interpretation, and intelligent exploitation of SAR imagery, this Special Issue is closely aligned with the scope of Remote Sensing, particularly the Remote Sensing Image Processing section.

Articles may address, but are not limited to, the following topics:

• SAR foundation models and large-scale pre-training;
• Transfer, adaptation, and generalization of SAR foundation models;
• Multimodal and multi-sensor foundation models for SAR;
• Vision–language models, multimodal large language models, and agentic SAR systems;
• Generative AI for SAR imaging and interpretation;
• Intelligent SAR image understanding and analysis;
• Physics-aware learning for PolSAR, InSAR, TomoSAR, and advanced SAR modalities;
• Large-scale SAR datasets, benchmarks, and evaluation methodologies;
• Efficient, explainable, robust, and trustworthy SAR AI;
• Foundation-model-enabled SAR applications in Earth observation.

Prof. Dr. Haipeng Wang
Dr. Lingjun Zhao
Guest Editors

Dr. Weijie Li
Guest Editors Assistant

Manuscript Submission Information

Manuscripts should be submitted online at www.mdpi.com by registering and logging in to this website. Once you are registered, click here to go to the submission form. Manuscripts can be submitted until the deadline. All submissions that pass pre-check are peer-reviewed. Accepted papers will be published continuously in the journal (as soon as accepted) and will be listed together on the special issue website. Research articles, review articles as well as short communications are invited. For planned papers, a title and short abstract (about 250 words) can be sent to the Editorial Office for assessment.

Submitted manuscripts should not have been published previously, nor be under consideration for publication elsewhere (except conference proceedings papers). All manuscripts are thoroughly refereed through a single-anonymized peer-review process. A guide for authors and other relevant information for submission of manuscripts is available on the Instructions for Authors page. Remote Sensing is an international peer-reviewed open access semimonthly journal published by MDPI.

Please visit the Instructions for Authors page before submitting a manuscript. The Article Processing Charge (APC) for publication in this open access journal is 2700 CHF (Swiss Francs). Submitted papers should be well formatted and use good English. Authors may use MDPI's English editing service prior to publication or during author revisions.

• synthetic aperture radar (SAR)
• foundation models
• deep learning
• self-supervised learning
• multimodal learning
• vision–language models
• generative AI
• intelligent SAR interpretation
• physics-aware learning